Using Past Incidents As Leading Indicators for Pipeline Integrity Management

Using Past Incidents as Leading Indicators for Pipeline Integrity Management

Hao Chen^*, Cynthia Spitzenberger, Julie Zhu, Lindsay Deal, and Tone Rice

* Contact Email: hao.chen@dnvgl.com

Section of Safety Risk Analysis, Department of Risk Advisory Services, DNV GL

1400 Ravello Drive, Katy, TX, U.S., 77449

Abstract:

PHMSA data indicates about 11,000 pipeline incidents have occurred in U.S. in the past 20 years, causing 400 fatalities, 1500 injuries, and over $6.5 billion property damages. Increasing energy demand urges the expansion of pipeline networks and their flexibility and capability to deliver various products. The concerns to satisfy the increasing energy demand together with increasingly stringent regulatory requirements impose significant pressure on pipeline operators to maintain safe operation. How to allocate limited resources to the sections most susceptible to failure still remains a big challenge to most operators. One feasible preventive measure is to learn from the past pipeline failures, in which failure data analysis is essential. From effective data analysis, useful failure information such as the pipeline materials, the involved failed parts, failure types, the most frequent failure causes, and the incident location environment, can be obtained, which can help increase the effectiveness of pipeline integrity management program by optimizing the allocation of limited resources to the “weak points” most likely tending to fail.

Although Europe-based pipeline incident databases such as EGIG and CONCAWE have been evaluated, it is beneficial to understand the U.S. pipeline failure data which more accurately reflects the industry situation in the country, especially for U.S. pipeline operators. Moreover, if some similarities between the pipeline failure patterns in U.S. and in Europe can be found, the experience of effective pipeline integrity management could be shared to enhance the safety performance of the pipeline industry in both regions. An extensive public pipeline incident database is accessible from PHMSA, but it only includes raw incident data and requires a systematic analysis to generate statistically meaningful results. In this work, the raw PHMSA data (particularly liquid pipeline data) are categorized and normalized across different periods, since the reported data format varies drastically over time. Then the normalized data are systematically analyzed to reveal the failure trends and causes. In addition, the failure patterns obtained for U.S. pipelines are compared with European data to highlight the similarities and differences, aiming to provide useful information for effective decision making on pipeline integrity management.

Keywords: 

Pipeline, Pipeline Incident, Pipeline Failure, Leading Indicator, Integrity Management